Modeling the mobility of mobile devices has always been a key issue for researchers as mobility needs to be taken into consideration in a variety of research situations in wireless networks. For many years, researchers have been striving to come up with more accurate mobility models which produce results similar to real mobility data. Although recent focuses on social relationship have improved the performance of mobility models, they still neglect or underestimate the influence of spatial and temporal factors which might greatly affect the movement of people. In this paper, we propose a novel way of modeling human movement on the basis of social, spatial and temporal factors by adopting a unique model structure and by introducing new mobility features such as roads. Simulation demonstrated that these measures produce results similar to real mobility data.